Research notes on every problem, and a library of the papers behind them. Built from the open erdos repository.
Updated
This source concerns signed sums and families of subsets. Its published-page scan contains a GDZ archive cover, then printed pages 251–259 on PDF pages 2–10. The mathematical version is the 1965 journal article.
Finite set notation
Write for all subsets of a finite set and when . In particular, . An antichain contains no two distinct comparable members.
A saturated symmetric chain in contains one set at each rank , for some , with successive sets related by inclusion. Its length is its number of members, , rather than its number of edges. The source calls these subchains of maximal chains. Empty lists arising in the induction are discarded.
Set for integers outside . For , write
These are the chain-tail counts proved at the remark after Lemma I. The number of antichains is a nonnegative integer. Its union is empty when ; the bound by largest rank levels is truncated at . All subset-family bounds use , permitting equality.
Signed sums and discs
For , the counted objects are sign choices for which lies in a region. Distinct choices giving the same complex number are counted separately. The unique empty choice at has sum zero.
A closed unit disc is ; an open unit disc is . The plane theorem uses and allows a closed disc. The finite scaling consequence uses and requires an open disc. The norm-one closed-disc version is false.
Inputs and coverage
The plane chain proves Lemma I, the chain-count remark, Lemma II, Theorem II, Theorem I, and the open-disc transfer. It includes the binomial identity used by Theorem II and uses only finite induction, finite counting, and elementary Euclidean inner-product and rotation facts. Although the source attributes Lemma II to Erdős 1945, its proof is included locally and is not an unproved imported input.
The later higher-dimensional branch has only stated source claims and proof pointers here. No complete proof of its Lemmas III/IV or Theorem III is claimed.
Bears on. Problem 498, with the explicit disc and multiplicity conventions above.